def load_from_cdx_test(cdx, revisit_func=load_orig_cdx, reraise=False, failed_files=None): resolve_loader = ResolvingLoader(test_warc_dir) cdx = CDXObject(cdx) try: (headers, stream) = resolve_loader.resolve_headers_and_payload(cdx, failed_files, revisit_func) print headers sys.stdout.write(stream.readline()) sys.stdout.write(stream.readline()) except ArchiveLoadFailed as e: if reraise: raise else: print 'Exception: ' + e.__class__.__name__
def _init_replay_view(self, config): cookie_maker = config.get('cookie_maker') record_loader = ArcWarcRecordLoader(cookie_maker=cookie_maker) paths = config.get('archive_paths') resolving_loader = ResolvingLoader(PathResolverMapper()(paths), record_loader=record_loader) return ReplayView(resolving_loader, config)
def _init_replay_view(self, config): cookie_maker = config.get('cookie_maker') record_loader = ArcWarcRecordLoader(cookie_maker=cookie_maker) paths = config.get('archive_paths') resolving_loader = ResolvingLoader(PathResolverMapper()(paths), record_loader=record_loader) redis_warc_resolver = config.get('redis_warc_resolver') if redis_warc_resolver: resolving_loader.path_resolvers.append(redis_warc_resolver) return WebRecReplayView(resolving_loader, config)
def __init__(self, query_handler, config=None): super(WBHandler, self).__init__(config) self.index_reader = query_handler cookie_maker = config.get('cookie_maker') record_loader = ArcWarcRecordLoader(cookie_maker=cookie_maker) paths = config.get('archive_paths') resolving_loader = ResolvingLoader(paths=paths, record_loader=record_loader) self.replay = ReplayView(resolving_loader, config) self.fallback_handler = None self.fallback_name = config.get('fallback')
def load_from_cdx_test(cdx, revisit_func=load_orig_cdx, reraise=False, failed_files=None): resolve_loader = ResolvingLoader(PathResolverMapper()(test_warc_dir)) cdx = CDXObject(cdx.encode('utf-8')) try: (headers, stream) = resolve_loader(cdx, failed_files, revisit_func) print(repr(headers)) sys.stdout.write(stream.readline().decode('utf-8')) sys.stdout.write(stream.readline().decode('utf-8')) except ArchiveLoadFailed as e: if reraise: raise else: print('Exception: ' + e.__class__.__name__)